fitonecycle!
function defined in module
FastAI
fitonecycle!(learner, nepochs[, lrmax])
Fit
learner for
nepochs using a one-cycle learning rate schedule. The learning rate starts at
lrmax/div for
pct_start*nepochs epochs, rising to
lrmax and then goes down to
lrmax/div_final over the remaining duration.
wd = 0: weight decay
pct_start = 0.25: Percentage of time spent raising the learning rate
div = 25: Starting learning rate is
lr_max/div
div_final = 1e5: Ending learning rate is
lr_max/div_final
There is
1
method for FastAI.fitonecycle!:
The following pages link back here:
Custom learning tasks, FastAI.jl, How to log to TensorBoard, How to train a model, Image segmentation, Introduction, Keypoint regression, Quickstart, Saving and loading models for inference, Siamese image similarity, Tabular Classification, TimeSeries Classification, Variational autoencoders, fastai API comparison, tsregression